v0.3.3

Revised algorithm for calculating relevance

1. The main factors influencing the relevance are taken into account:

   • number of words found in a text (influence updated in this release)
   • abundance of words in all indexed texts (frequent words have less influence on relevance)
   • proximity of different found words from a query to each other (influence updated in this release)

2. Processing of words from titles and from keywords was refined. Now titles are more important, and there is no double counting of relevance from repeating words in both title and keywords.

3. The debug call $resultSet->getTrace() contains more information about the influence of each factor in the overall relevance.

Other features

• Implemented highlighting of a whole phrase for adjacent search words (closes #9).
• Added new method ResultSet::getTotalCount() that can be useful for building pagination links (page 1, page 2, ... page N).
• Added new method PdoStorage::getIndexStat() for the disk usage of index tables in database.
• Refinements in Russian stemmer.
• Improved html cleanup for snippets (contents of <style> and <script> tags are now being removed).

v0.3.2

Moving to github actions.

v0.3.1

Fixed issues of unique word index and utf8 collations.

v0.3.0: Merge pull request #16 from parpalak/instances

Implemented dividing search index into parts (instances) that can be used independently

v0.2.5

Added full support for multi-language morpology.

v0.2.4

Another fix for utf8mb4 support.

v0.2.3

Fixed tests for PHP 7.2 and 7.3.

v0.2.2

Fixed utf8mb4 support.

v0.2.1

Fixed XSS when highlighting text with correctly encoded HTML entities.

v0.2.0

Removed TOC caching and fetching all the TOC entries.